TRANSIENT HOLES IN THE ERYTHROCYTE MEMBRANE …

T R A N S I E N T H O L E S I N T H E E R Y T H R O C Y T E

M E M B R A N E D U R I N G H Y P O T O N I C H E M O L Y S I S

A N D S T A B L E H O L E S I N T H E M E M B R A N E A F T E R

L Y S I S B Y S A P O N I N A N D L Y S O L E C I T H I N

P H I L I P S E E M A N

From The Rockefeller University, New York. Dr. Seeman's present address is the Department of Pharmacology, Cambridge University, England

A B S T R A C T

Ferritin and colloidal gold were found to permeate human erythrocytes during rapid or gradual hypotonic hemolysis. Only hemolysed cells contained these particles; adjacent intact cells did not contain the tracers. Ferritin or gold added 3 min after the onset of hypotonic hemolysis did not permeate the ghost cells which had, therefore, become transiently permeable. By adding ferritin at various times after the onset of hemolysis, it was determined that for the majority of the cells the permeable state (or interval between the time of development and closure of membrane holes) existed only from about 15 to 25 sec after the onset of hemolysis. It was possible to fix the transient "holes" in the open position by adding glutaraldehyde only between l0 and 20 sec after the onset of hemolysis. The existence of such fixed holes was shown by the cell entry of ferritin and gold which were added to these prefixed cells. Membrane defects or discontinuities (of the order of 200-500 A wide) were observed only in prefixed cells which were permeated by ferritin subsequently added. Ad- jacent prefixed cells which did not become permeated by added ferritin did not reveal any membrane discontinuities. Glutaraldehyde does not per se induce or create such membrane defects since cells which had been fixed by glutaraldehyde before the 10-see time point or after the 180-see time point were never permeable to added ferritin, and the cell membranes never contained any defects. It was also observed that early in hemolysis (7-12 sec) a small bulge in one zone of the membrane often occurred. Ghost cells produced by holothurin A (a saponin) and fixed by glutaraldehyde became permeated by ferritin subsequently added, but no membrane discontinuities were s e e n . Ghosts produced by lysolecithin and fixed by glutaraldehyde also became permeated by subsequently added ferritin, and many membrane defects were seen here (about 300 A wide).

I N T R O D U C T I O N

The work in this paper is concerned with the disruption and repair of the erythrocyte cell membrane during hypotonic hemolysis and with the time course of these events. The ultrastructure of the membrane defects involved in osmotic hemolysis and in the hemolysis induced by two surface-

active agents, lysolecithin and holothurin A, is also described.

H Y P O T O N I C H E M O L Y S I S : After hypotonic hemolysis the cell subsists as a distinct cellular corpuscle (1) and retains osmotic properties char- acteristic of semipermeable membranes (2--4). The

55

experiments of TeoreU (4) and of Hoffman (5) demonstrated that erythrocytes hemolysed in 5 parts water (such that the final concentration of the hemolysate was 20 mM Na and K; Teorell, reference 4) or in 9 parts of 18 mM NaCI (5) spontaneously recover their low permeability to

Na, K, glucose, and sucrose. While the untreated ghost can be impermeable to

low molecular weight substances, it is known that during the course of hypotonic hemolysis the erythrocyte becomes highly permeable to substances of low and high molecular weight (6-8). Some time after the completion of hemolysis, however, the cells recover their normally low permeability. Exceptions to this impermeabili ty of ghosts have been reported (9, 4, 5).

The size and number of holes that develop during hemolysis are not known. Various estimates for the size have been made and range from less than 100 A in diameter (i0) to over 100 A (9) and even up to diameters of the order of 1 # (11, 12). In order to check on the size of the hole, ferritin (molecular weight of around 106; diameter of about 120 A total; diameter of electron- opaque core about 80 or 90 A; Haggis, reference 13) and colloidal gold (25-300 A diameter) were used in the experiments to be described.

L Y T I C H E M O L Y S I S : Saponin may produce holes in the erythrocyte membrane (14-17). The work reported in the present paper indicates that there are stable holes or defects in saponin- and lysolecithin-treated erythrocytes.

M A T E R I A L S A N D M E T H O D S

Collection of Blood

The human erythrocytes were obtained from a fasting human volunteer and the blood was heparinized (100 units per milliliter blood).

Preparation of Gold Suspensions and Ferritin Solutions

Horse spleen ferritin was obtained from Nutritional Biochemieals Corp., Cleveland, Ohio, as cadmium- free, twice-crystallized ferritin. Colloidal gold suspension was obtained from Abbott Laboratories. Chicago. These stock solutions of gold and ferritin were then dialysed for 48-72 hr against a dialysing medium about 40 times the volume of the dialysate; this medium was changed every 12 hr. The composi- tion of the dialysing medium was either 154 mM NaC1 or 25 mM NaC1 in l0 mM sodium phosphate buffer, pH 7.0. The hypotonic ferritin plus gold solution was pre-

pared by dialysing the stock ferritin and gold suspensions individually against 25 rnM NaC1 (pH 7) and mixing 1 volume of each just before the experiment.

Hppotonic Hemolysis Experiments Involving Ferritin and Gold

1. S T A N D A R D R A P I D H E M O L Y S I S P R O C E D U R E :

An aliquot of erythrocytes (either 0.02 ml of packed cells or 0.04- ml of whole heparinized blood) was added to either 0.5 ml o1" 1.0 ml of 10% ferritin (and/ or colloidal gold each at half the stock concentration) containing 25 mM NaC1 in 10ram sodium phosphate buffer, pH 7.0. The cells were mixed by means of a vortex mixer and after 5 min they were fixed in suspension by the addition of 2 or 3 ml of either 2% osmium tetroxide (Palade's fixative; reference 18) or 2% glutaraldehyde (in 0.1 ~ sodium cacodyiate- HC1, pH 7.2). Erythrocytes that had been fixed only in 2% OsO4 were exposed to the OsO4 no longer than 20 rain total time. This short fixation time was chosen to preclude or diminish the possibility of artifactual translocation of ferritin or gold across the cell membrane. (The definition of, and the problem of, artifactual translocation is presented in the Discussion.) The duration of glutaraldehyde fixation was from 20 to 30 min, which includes the time required for centrifugation. After the cells were fixed in suspension with glutaraldehyde, the erythrocytes were centrifuged at about 11,000 g for 8-10 min to form a pellet. These glutaraldehyde-fixed pellets of cells were then fixed in 2% OsO4 for 30 rain to 2 hr; it was found that after glutaraldehyde fixation such long periods of OsO4 fixation did not cause any artifactual translocation of particles. The pellets were then gently broken into l-ram pieces and in most experiments were treated for 1 hr in 0.5% uranyl acetate in Ver- onal-acetate-HC1 buffer (19). This uranyl treatment did not cause artifactual translocation of particles in glutaraldehyde-OsO4-fixed cells, and greatly en- hanced the contrast of the membrane. The pellets were dehydrated, embedded in Epon 812, and sectioned (silver sections) ; the sections were stained with uranyl and lead (19). The lead citrate stain causes considerable loss in contrast of the fcrritin particles and was, therefore, kept to 1 rain or less. Much higher ferritin contrast is obtained by following the uranyl with the Karnovsky stain (20), and many sections were stained in this way. The sections were examined in the RCA-EMU 3F, the Hitachi HS-7S and the Siemens Elmiskop I electron microscopes.

Individual colloidal gold particles were readily identified by their circular profile (95-300 A in diameter) and their extreme electron opacity. Ferritin particles were also easily identifiable when the sections were stained with uranyl and Karnovsky B stains. In agreement with Haggis (13), ferritin was observed and photographed as circular and usually uniformly

56 THE JOURNAL OF CELL BIOLOGY • VOLUME 3~, 1967

dense-cored particles about 90 A in diameter . As ment ioned above, ferritin in sections s tained with lead citrate had less electron opacity. I n these sections the density of ferritin approached the low density of the i rregular and dispersed cytoplasm. T he dist inction between ferritin and isolated pieces and strands of cytoplasm was, however, vir tually never in doubt . Firstly, wi th the exception of the chemical ly fixed hole experiments , there were usual ly hundreds of intracellular particles all about 90 A in diameter . No erythrocyte hemolysed in the absence of ferritin ever exhibi ted un i fo rm dense dots of this sort. Second, a rout ine procedure was to make several mic rograph prints of each microscope plate, one of which was a h igh contrast and underdeve loped print. This pr in t would usual ly show up the existence of ferritin to advan tage while suppressing the low density cytoplasmic remnants .

2. PROCEDURE OF GRADUAL HEMOLYSIS BY DIALYSIS : An al iquot of whole blood (0.04 ml) was added to 1.0 ml of fcrritin (at half the stock concentration) in 154 mM NaC1, p H 7, in a dialysis bag. The th in dialysis bag was then dialysed against 250 ml of 25 mM NaC1, p H 7, for 20 rain, dur ing which t ime the tu rb id crythrocyte suspension cleared completely; clearing occurred at about 6-8 mi n after the onset of dialysis. An aliquot (0.05 ml) of 2.3 M NaC1 was then added and the cells were fixed in suspension wi th Palade 's fixative (18). T h e control for this exper iment was as follows. Erythrocytes were added to 0.5 ml of 154 mM NaCI and dialysed for 10 min . After tha t t ime, 0.5 ml ferritin in 25 mM NaC1 was added and dialysis carr ied on for another 10 rain; 2.3 M NaC1 (0.05 ml) was then added and followed by Palade 's fixative (18).

3. TIME COURSE OF THE DEVELOPMENT OF HOLES: An al iquot of 0.02 ml packed cells was pipet ted into the bo t tom of a test t ube ; 1 ml of ferritin (at ha l f the stock concentrat ion) in 25 mM NaC1 in 10 mM sod ium phospha te buffer, p H 7, was added for vary ing periods of t ime (5, 10, 20, 30, and 60 sec); after these t ime periods, 0.05 rnl of 2.3 M NaC1 was added (to make the final solution isotonic and to close the p resumed holes) for corresponding t ime periods (5 min , 55 sec; 5 min , 50 sec; 5 min , 40 sec; 5 min , 30 sec; and 5 min , respectively). T h e erythrocytes, now hav ing been exposed to ferritin for a total t ime of 6 rain, were fixed in suspension wi th 1 ml of 2 % OsO4 (Palade's fixative m a d e isotonic wi th NaCI instead of with sucrose) for 5 rain. Photographs of the erythrocytes in th in section were t aken in the R C A - E M U - 3 F microscope. Between 50 and 70 cells for each experi- menta l t ime point were r andoml y photographed.

4. TIME COURSE OF THE CLOSING OF HOLES: An al iquot of 0.09 ml packed cells was pipet ted into the bo t tom of a test tube and 0.5 ml of 25 mM NaC1 (pH 7) was added for vary ing periods of t ime (15, 30, 60, 120, and 180 sec) ; 0.5 rnl of fcrritin in 25 mM NaC1 (pH 7) w a s t h e n added for 3 rain. A n al iquot (0.05 ml)

of 2.3 M NaC1 was added to make the cell suspension isotonic and, after 3 min , 1 rnl of 2 % OsO4 (Palade's fixative m a d e isotonic with NaC1) was added to the suspension for 5 min . T h e ferritin, therefore, was in contact wi th the erythrocytes for 11 rain in all the tubes. Approx imate ly 50 cells for each t ime point were pho tographed in the electron microscope.

5. THE FIXATION OF THE TRANSIENT HOLES IN THE OPEN POSITION: T h e procedure was as follows. An al iquot (0.04 ml) of whole blood was pipet ted into the bo t tom of a test tube and 1 ml of 25 mM NaC1 (pH 7) was added to start hypotonic hemolysis; after 7, 12, 18, 22, 60, and 300 sec, 3 ml of 2 % glu ta ra ldehyde were added. T h e fixed cells were t hen centr i fuged (10,300 g for 10 rain at 18°C) into a pellet which was washed twice in Michael is buffer (18). T h e cells were finally resuspended in 0.5 ml ferritin or ferritin plus gold in 154 mM NaC1 for 30 rain; the cells were then centr i fuged into a pellet once more and OsO4 was added.

6. EFFECT OF SOLUBLE HEMOGLOBIN ON FER- RITIN ENTRY INTO CELLS: T o t e s t t h e p~ssibil- ity tha t the released hemoglobin adsorbs and thereby restricts the entry of ferritin, the following experi- menta l procedure was carr ied out. An al iquot (0.08 ml) of whole blood was added to 1 ml of 25 mM NaC1 (pH 7) and, after 5 rain, the ghosts were centr ifuged down at 10,000 g for 10 rain; the hemoglobin-conta in ing supe rna tan t was saved and the ghosts were discarded. An al iquot (0.04 ml) of whole blood was now added to 1 ml of ferr i t in-hemoglobin solution (0.5 ml 10% ferritin in 25 mM NaC1 plus 0.5 ml of the hemoglobin in 25 mM NaC1) and, after 5 rain, an al iquot (0.05 ml) of 2.3 M NaCl was added (to restore isotonicity) followed by the addi t ion of 2 ml of Pal- ade's fixative (18).

7. HEMOLYSIS BY LYTIC AGENTS: Holo thur in A (a saponin prepara t ion purified from the sea cu- cumber , Actinopyga agassizi) was kindly dona ted by Dr. J . D. Chanley, the M o u n t Sinai Hospital , New York. Lysolecithin was obta ined from K & K laboratories, J a m a i c a , N. Y. T h e exper iments involving hemolysis by lytic agents were carr ied out in the following manner . An al iquot (0.04 ml) of hepar in ized whole blood was added to 1 ml of 154 mM NaC1, p H 7, and the cells were hemolysed by the addi t ion of 30/zg of holo thur in A (in 0.05 ml of 154 mM NaC1) or 0.2 m l of 3 X 10 -8 M lysolecithin. T h e turbid erythrocyte suspension cleared at a round 5 to 6 rain in the presence of holo thur in and at a round 14 rain with lysolecithin. 10 rain after clearing, 3 ml of g lu ta ra ldehyde fixative were added for 5 m in and the cells were centr ifuged at 10,300 g for 14 min. The pellets were washed twice in 154 mM NaCI, p H 7, (without resus- pending the cells) and the cells finally resuspended in 0.5 ml ferritin in 154 mM NaC1, p H 7, for 5 rain. T h e cells were then fixed in Palade 's fixative 08 ) .

PHILIP SEEMAN Transient and Stable Hole.~ in Erythrocyte Membrane 57

R E S U L T S

Entry of Ferritin into Erythrocytes during Hemolysis

Ferritin was found inside erythrocytes which had been hypotonically hemolysed in the presence of ferritin (Methods, part 1). This is shown in Fig. 1 a. The ferritin appears as extremely dense dots about 90 A in diameter when the thin sections have been stained by means of Karnovsky B lead stain. These black ferritin dots are seen against the irregular gray background of residual intracellular material. Fig. 1 a also shows the corner of an erythrocyte which did not undergo hemolysis, as indicated by the homogeneous dense cytoplasm; such nonhemolysed cells never contained any ferritin particles. Many hundreds of cells were examined during the course of 25 experiments and every hemolysed cell contained ferritin. Varying the osmotic pressure or the pH of the solution did not affect the location of ferritin, other than change the proportion of hemolysed cells.

The relationship between per cent hemolysis and the per cent of cells containing ferritin was also studied. An osmotic fragility curve was obtained in the usual way by adding erythrocyte aliquots to a series of varying hypotonic NaCl solutions and measuring the hemoglobin released into the cell-free supernate at 543 m/~. A duplicate series of hypotonic solutions containing ferritin was treated in the same way, except that the cells were fixed and examined for intracellular ferritin (Methods, part 1). (A duplicate series was neces-

sary since the intense optical density of ferritin interfered with the spectrophotometry of hemoglobin.) I t was found that the proportion of cells containing ferritin exactly paralleled the per cent hemolysis.

No membrane discontinuities were observed in these cells (which had been fixed 5 min after the onset of hemolysis). A sample of the intact membrane is shown in Fig. 1 b; here the middle part of the photograph is the intracellular space and the ferritin dots appear very gray with low contrast (since the lead citrate stain was employed), The portion of the cell shown in Fig. 1 b, although demonstrating the intact membranes to advantage, does not contain abundant numbers of ferritin particles in the thin cytoplasmic region; this particular cell does, however, contain hundreds of particles in the wider expanses of cytoplasm situ- ated 1-2 p above and below the selected zone

shown.

Entry of Colloidal Gold into Erythrocytes during Hemolysis

Colloidal gold particles (25-300 A in diameter) were also found to enter erythrocytes which had been hypotonically hemolysed in the presence of the gold. This is shown in Fig. 1 c and d, where the ferritin molecules appear gray and the gold dense black.

The Impermeability of Erythrocyte Ghosts to Ferritin or Gold Particles

When ferritin or ferritin plus gold in hypotonic solution was added 5 rain after the onset of hemol-

FIGURE I Showing that ferritin and colloidal gold enter erythrocytes during the 5-rain period of hypotonic hemolysis.

FIGURE 1 a The cells were hypotonically hemolyzed in the presence of ferritin and then fixed in 2% OsO4. The hemolyzed cell contains fcrritin (seen as black dots, 90 A wide) while the intact cell does not. RCA-EMU-3F. Uranyl and Karnovsky B stains. X 41,000.

I~IGVRE 1 b The cells, hemolyzed in the presence of ferritin (stained as pale gray dots here), were fixed in glutaraldehyde, then OsOa, and treated with uranyl acetate. Perritin is inside (center of photo) and the unit membranes are not disrupted. Elmiskop I. Urany] and lead citrate stains. X 1~8,000.

FIGURE 1 C Same as Fig. 1 b, except that the cells were hemolyzed in the presence of both colloidal gold (black dots) and ferritin (gray dots). X 6%099.

FIGURE 1 d Showing continuity of the membrane in a cell hemolyzed in the presence of gold (black dots) and ferritin (gray dots). Procedure as in Fig. 1 b. X ~06,000.

58 TaE JOURNAL OF CELL BIOLOGY • VOLUME 3~, 1967

PHILIP SEEMAN Transient and 8table Holes in Erythrocyte Membrane 59

ysis, then no particles entered the cells. Examples of such results are shown in Fig. 2 a and b. Re- storing tonicity by the addition of an aliquot of hypertonic NaC1 solution to make a final concentration of 150 mM before adding the ferritin also gave the same results.

There are two possible interpretations of the fact that no ferritin was found inside the erythrocyte when the ferritin was added 5 rain after the onset of hypotonic hemolysis. One possibility is that the holes have spontaneously closed and that the ghost membrane has become impermeable to the ferritin.

A second conceivable situation is as follows. When ferritin is added 5 min after the onset of hemolysis, the ferritin may become quickly and totally adsorbed to the abundant hemoglobin which has been released to the extracellular space during the 5-rain period. Such adsorption would markedly reduce the extracellular concentration of free ferritin particles. In this case, virtually no ferritin would enter the erythrocyte ghosts even if the holes in the membrane still existed. In this event, therefore, the absence of intracellular ferritin would not indicate anything about the permeability of the membrane to ferritin. To test this second possibility, erythrocytes were hemolysed in the presence of ferritin and hemoglobin (Meth- ods, part 6). By electron microscopy it was observed that all the hemolysed cells contained ferritin. The presence of extracellular hemoglobin, therefore, does not prevent the permeation of ferritin into the cell.

Gradual Hemolysis

Erythrocytes were hemolysed in the presence of ferritin by gradual hemolysis (references 1 l, 21; Methods, part 2); all the hemolysed cells con-

tained ferritin. None of the erythrocyte ghost control cells contained any ferritin.

Gradual lysis was also done in the presence of

colloidal gold. Gold particles less than 150 A in diameter were found inside the cell; although larger particles were not found in the cytoplasm it

was not possible to say whether these large particles were excluded because of their size or because there were simply not enough particles greater than 200 A in diameter.

Time Course of the Development and Closing of Holes in Hypotonic Hemolysis

The time course of the development and closing of membrane holes was investigated as follows.

l. THE DEVELOPMENT OF HOLES: The ra-

tionale here was to time the development of the holes by inducing hypotonic hemolysis in the presence of ferritin and by restoring the isotonicity of the medium (thereby presumably leading to a rapid closure of the holes) at short intervals there- after (Methods, part 3). Only the hemolysed cells (with mottled intraceUular texture) contained ferritin. The number of cells containing ferritin was counted and plotted as a per cent of the total number of approximately 60 cells for that t ime point. The result is shown in Fig. 3, curve A. I t is seen that 50 % of the cells has been permeated by ferritin at around 15 sec. At 60 sec, all the cells have been permeated. Curve A represents the time course of the development of the holes or pores, therefore. I t should be mentioned that the location of ferritin at the time points studied here was virtually an all-or-none phenomenon; that is, either there was not one single particle to be found in the hemolysed cell or else the cell was richly loaded with ferritin.

FIGURE ~ The impermeability of erythrocyte ghosts to ferritin or gold particles. Erythro- cytes were first hypotonieally hemolysed for 5 rain in the absence of ferritin or colloidal gold.

FIGURE ~ a After hemolysis the medium was made isotonic with NaC1, OsOa was added, and then ferritin. No ferritin (black dots) permeated the fixed ghosts. Identical results were obtained if the ferritin was added before the NaC1 or before the OsO4. RCA-EMU- 3F. Uranyl and Karnovsky B stains. )< 44,000.

FIGURE ~ b After hemolysis ferritin (pale gray dots) and gold (black dots) in hypotonic solution were added for another 5 min befo1~e fixing in glutaraldehyde (followed by OsOa and uranyl solutions). No cytoplasmic particles are seen. Elmiskop I. Uranyl and lead citrate stains. X 9e,000.

60 THE JOURNAL OF CELL BIOLOGY " V O L U M E 3 ~ , 1967

PIIILIP SEEMAN Transient and Stable Holes in Erythrocyte Membrane 61

Time course of the opening and closing of "pores"

,oo = °'-..\ E

A. Ferritin odded with ~ . J J-B. Hypotonic sol'n ._c~ hypotonic sol'n ~ . , / ~ ~ added before ferritin

, .> , . , v f s 4 / \k50% o, ce, 50 50% of cell Is

o o a r e a l l \ ~ r~ closing

o . j \ o .

°~ J - ~ ' / / . ~ . ~ . o o .Z~., , i . . . . I ~ , , i , , _ I . . . . J . . . . I , ,// ~ - - - - T - - ' ~ " ' ~ a - ~ _ 0 5 tO 15 20 25 50 I 2 5 4 5

Seconds Minules

FIGURE 3 Data delineating the transient permeable state of the hemolysing erythrocyte. A Cell-opening series The cells were placed in hypotonic solution containing ferritin, and after different time intervals NaC1 was added to make the medium isotonic, The cells were fixed in OsO4. Each point represents the fraction (of around 60 cells) which contain ferritin. 50% of the cells has admitted ferritin at 15 sec. B Cell-closing series The cells were put into hypotonic solution not containing ferritin; at varying times after the onset of hypotonic hemolysis ferritin was added and the cells were later fixed in OsO4. Most of the cells have closed at 80 sec. Considering A and B together, the majority of the cells have "holes" or "pores" existing for a 10-scc interval between 15 and o~5 sec.

2. T H E C L O S I N G OF H O L E S : T he rat ionale here was to add ferritin at various times after the onset of hemolysis (Methods, par t 4); since the holes are known to close spontaneously (Fig. 2 b), ferritin should not enter if it is added late in the hemolysis period. Al though all the cells were hemolysed, not all conta ined ferritin. The proport ion of cells conta ining ferritin at each t ime point is g raphed in Fig. 3, curve B. I t is seen that , when the ferr idn was added at 15 sec, all the cells picked up the ferritin. Only 18~0 of the cells was permeated by ferritin if the lat ter was added 30 sec after the onset of hemolysis. This indicates tha t most of the cells have closed at 30 sec. Interpolat ing, it can be seen tha t 50 % of the cells is spontaneously closing at a round 25 sec. Ferr i t in added 1, 3, and 5 rain after the onset of hemolysis did not enter the ghosts; in one experiment, however, it was observed that ferritin added 3 rain after the onset of hemolysis entered abou t 10% of the erythrocytes tha t were hemolysed. This is indicated in Fig. 3, curve B.

I t is interesting to note that , among those cells

examined at the 15-see t ime point, two were hemolysed but did not contain any ferritin. This

indicates tha t these two cells hemolysed and closed

in the 15-see interval before the ferritin was added.

15 sec, therefore, represents the upper l imit for the

durat ion of the existence of the t ransient holes of these two cells.

3. T H E F I X A T I O N OF T R A N S I E N T H O L E S IN

THE OPEN POSITION: The results shown in Fig. 3, curve A, indicate tha t 50 % of the cells is developing holes at a round 14 or 15 sec; Fig. 3, curve B, indicates tha t 50% is closing between roughly 24 and 25 sec. These da ta imply tha t the majori ty of the cells have holes at abou t 19 or 20 sec and tha t the holes remain open for abou t 10 sec. I t was necessary to check these two points by some different procedure. In the hope of fixing the t ransient holes in the open position and hav ing ferritin enter the chemically-fixed holes, OsO4 or glutaraldehyde fixative was added to the cell suspension at various t ime points dur ing hypotonic hemolysis, and this was followed by the addition of ferritin. Fixation by OsO4 was tried with bo th isotonic and hypotonic fixatives added at 21 sec after the onset of hemolysis; a l though many if not all of the cells were hemolysed, none conta ined ferritin. The OsO4 apparent ly failed to fix the t ransient holes in the open position or to keep them open long enough for ferritin to be admit ted.

I t proved possible, however, to fix the t ransient

holes in the open position by means of glutaralde-

hyde (Methods, par t 5). Ferri t in and gold added

to cells which have been fixed dur ing the first

62 T E E JOURNAL OF CELL BIOLOGY • VOLUME 3~, 1967

minute of hemolysis are found to permeate many of the cells (see Fig. 4). Approximately 70 cells were photographed for each time point and the proportion of cells containing ferritin was plotted as a per cent of these 70 cells. In Fig. 5 it is seen that the majority of the cells are open at around 15 see, and this agrees fairly well with the value of about 19 or 20 sec obtained in the hole-development and hole-closure series of experiments. The time interval between the ascending and descending limbs of Fig. 5 is about 10 sec (at the 50% to 90% level of the ordinate), and this represents the duration of the existence of the holes for the majority of the cells. At 60 sec, there are a number of cells (8%) which have persisting holes or holes which have developed late in the hemolysis period (see Fig. 4 b).

Anatomy of the Membrane Defects or Holes

Developed during Hypotonic Hemolysis

A large number of cell profiles in the hole-fixation series revealed membrane breaks or discontinuities. Three examples are shown in Fig. 6. The membrane defects are of the order of 200-500 A wide. I t seems that the leaflets of the unit membrane at the edge of the defect end abruptly and do not join one another (except at the double-shafted arrow in Fig. 6 a--see legend to Fig. 6 and Dis- cussion). Only those cells which contained ferritin exhibited membrane defects. Adjacent ghosts, free of ferritin, had intact membranes (Figs. 4 b and 6 b and c). The defects were not found all around the perimeter of the cell profile but were clustered in one zone (about 1 # long), for any particular cell profile. Two clusters of defects in different parts of the same cell membrane profile were not observed, although in Fig. 6 b a profile is shown where two

defects are separated by about 1 #. I t is interesting to note that in these experiments the cytoplasmic

ferritin was often seen to be located near the defect zones. For example, ferritin is seen near the holes in the cell in Fig. 6 b but the rest of the cytoplasm (not shown) was entirely devoid of

ferritin.

Bulging of the Cell Membrane before the Development of Membrane Holes

Many of the cells fixed at 12 sec contained small patches of membrane which ballooned out (Fig. 7). Such profiles were also seen at 7 sec but were most numerous at 12 sec and nonexistent at later times.

Such ballooning may represent the precursor state of the membrane before the defects appear, since there was only one such bulge per cell profile and since these cells never had any ferrltin in them.

Stable Defects in the Membrane after

Lysis by Surface-Active Agents

While transient holes exist during hypotonic hemolysis, it was found that there were holes or membrane defects persisting long after hemolysis (10 rain) by holothurin A (a saponin) and lysolecithin. The results with lysolecithin (see Methods, part 7) are shown in Fig. 8. The cell contour is irregular and scalloped; ferritin is found inside the cells. At the base of the indentations caused by lysolecithin there seems to be a discontinuity of the membrane (Fig. 8 b). The results with the saponin are almost identical insofar as it creates ghosts which have scalloped contours and which are penetrated by subsequently added ferritin. The membrane pits or indentations created by the saponin are, however, only about one-fourth as deep and the membrane was always observed as continuous, even at the base of the pits; this continuity was seen despite the fact that ferritin permeated these glutaraldehyde-fixed ghosts.

Membrane Thickness

The mean value for 40 ghost membranes was 71 A (cf. references 22-25). The measurements were made on photographic enlargements and refer to the over-all distance from edge to edge of the unit membrane, the edges of which were judged by eye.

D I S C U S S I O N

The general observation of this study is that erythrocytes become momentari ly permeable to

ferritin and colloidal gold during hypotonic hemolysis, but persistently permeable after lytic hemol-

ysis.

Artifactual Translocation of Ferritin

The intracellular location of ferritin, added in isotonic solution to the extracellular space, is gen- erally considered to result from a permeation of ferritin during the process of fixation, embedding, and staining, and this process is commonly re- ferred to as artifactual translocation of ferritin. Such artifactual translocation of ferritin is induced by prolonged OsO4 fixation and has been seen by

PHILIP SEEMAN Transient and Stable Holes in Erythroeyte Membrane 63

some investigators (26-30) bu t not by others (31). No cytoplasmic free ferritin was noticed by Huxley (32) who used 2 hr of glutaraldehyde fixation (followed by OsO4).

U n d e r the conditions employed in the present series of experiments there was no evidence of ar t i factual translocation of ferritin into the cell. Ferr i t in added in isotonic solution to cells which were subsequently fixed and t reated as outlined in Methods never permeated any of the erythrocytes. Fur the r exper imenta l support for the absence of art ifactual translocation is shown by the ferritin- free cells in Figs. 1 a, 2 a, 2 b, 4 a, 4 b, 6 b, and 6 c as well as by those ferritin-free cells observed in other control experiments reported in Results (see sections on gradual hemolysis, and on the unsuccess- ful a t t empt of osmium tetroxide fixation of transient holes in the open position).

The C~ncentration and Location of Tracer

A comparison of all the micrographs indicates tha t the final extracellular concentra t ion of ferrifin or gold is highly variable. The reason for this is unknown. Presumably these tracers are extracted in varying degrees depending on the n u m b e r of changes of OsOa fixative (usually 2 changes when the cells had been prefixed in glutaraldehyde) , of uranyl solution (1 or 2 changes), and of dehydra t - ing solutions (5 to 7 changes of absolute ethanol and 4 to 6 changes of propylene oxide). A second point is tha t ferri t in is usually but not invar iably associated with dense material . This may represent adsorbed ferritin which is not easily extracted.

In the experiments on hole-fixation (Figs. 4, 5, and 6), wherein ferrifin was added to glutaraldehyde-fixed cells, i t was observed tha t in most cases the intracel lular presence of ferritin was all-or-none

(Fig. 4) as it was in the experiments of Fig. 1. In other experiments it was observed that ferrit in was not uniformly dis tr ibuted th roughout the cell bu t was located near the zones of holes. W h e t h e r a cell wi th chemically fixed holes contains few or m a n y particles presumably depends, among other factors, on (a) the m o m e n t selected for fixation (since this determines the size and n u m b e r of holes), (b) the density and viscosity of the fixed cytoplasm through which the ferritin must diffuse, and (c) the length of t ime which the cells have been exposed to ferritin.

The Time Course of Development and

Closure of the Transient Holes

In compar ing the two sets of results where in the t ime course of the deve lopment and closure of the holes was examined (Figs. 3 and 5), it is seen tha t there seems to be a discrepancy of a round 5 sec or so. In Fig. 3 the m a x i m u m rate of rise is a t 15 sec whereas in Fig. 5 it is a t about 10 sec; likewise, the m a x i m u m rates of decline differ by 5 or 10 sec. The discrepancy is not serious if one considers tha t the procedures are vastly different and tha t the pipet t ing and mixing accuracy are no bet ter than plus or minus 2 or 3 sec. I t should be pointed out

that , a l though the absolute values of the abscissae

of Figs. 3 and 5 may be compared, there is no

mean ing in a comparison of the absolute values of

the ordinates.

Gradual Hemolysis Compared to

Rapid Hemolysis

I t has been shown by Katchalsky et al. (21) tha t

erythrocytes undergoing a gradual hemolysis (1 l)

hemolyse a t lower extracellular tonicities com-

FIGUaE 4 Demonstrating the penetration of ferritin and colloidal gold into cells which have been fixed in glutaraldehyde at ~ and 60 sec after the onset of hypotonic hemolysis.

FIGURE 4 a At ~ sec after the cells were placed in hypotonie solution free of ferritin, the cells were fixed in glutaraldehyde, centrifuged, washed, and exposed to ferritin (seen here as black dots) which permeated 15% of the cells in the field (see Fig. 5); the ceils were then fixed in OsO4 but not given the uranyl treatment. RCA-EMU-SF. Uranyl and Karnovsky stains. X 35,000.

FIGURE 4 b Same as in Fig. 4 a, except that both ferritin (gray dots) and gold (black dots) were added to cells which had been fixed at 60 see. This cell is unusual, in that either the "holes" have persisted for a long time (over 45 sec) or the cell has hemolysed late. Uranyl treatment was given after the fixation. Elmiskop I. Uranyl and lead citrate stains. )< 160,000.

64 T H E JOURNAL OF CELL BIOLOGY - VOLUME 3¢~, 1967

I~XLn' S E E ~ Transient and Stable Holes in Erythroeyte Membrane 65

4O

~3o 'E

o 20

0

Entry of ferrifin inlo cells fixed during hemolysis

I I I I I I I l l l l l l l l l l ] i i I i i I l I I I I

2 5 I0 20 50 I00 200

Seconds

FIGURE 5 Showing the time course of the permeable state of ilemolysing erythrocytes to ferritin. At various times after the onset of hypotonic hemolysis in a solution free of ferritin, glutaraldehyde was added, and the cells were centrifuged and exposed to ferritin; the cells were then fixed in OsO4. Each point represents tile fraction of cells (out of 70 cells) COR- taining ferritin. I t is seen that the majority of cells are patent between 10 and 20 see after the onset of hemolysis. There are some cel]s (at 60 sec) which have either persisting holes or holes that have developed late in hemolysis.

pared to the rout ine rapid hemolysis; Katchalsky et al. a t t r ibuted this difference to a more gentle stretching of the cell m e m b r a n e by the dialysis procedure. I t was conceivable, therefore, tha t erythrocytes hemolysed by means of the gradual hemolysis technique would not contain ferritin. The fact tha t ferritin does permeate into ceils du r ing gradual hemolysis indicates tha t the stretch is not so gentle as to sieve out ferritin. I t will be necessary to fix cells a t various times dur ing gradual hemolysis in order to check the existence of m e m b r a n e slits, holes, or pores.

Anatomy of the Membrane Openings

I t is possible tha t the discontinuities or openings in the cell m e m b r a n e profiles shown in Fig. 6 are all really pa r t of an i r regular star-shaped defect and tha t the th in section has merely passed th rough the points of the star-shaped lesion. This possibility is only one of other conceivable geometrical con- figurations, the existence of which will have to be examined by other means and techniques. Useful

information would come from study of these mem- b rane defects by (a) stereoscopic electron microscopy, (b) negative staining techniques, (c) serial section microscopy, and (d) pho tograph ing areas of the cell wherein the m e m b r a n e has been grazed in section. A second point is tha t whereas the cell openings measure a round 300 A in wid th they must also be. a t least 500 or 600 A in depth, for this is the thickness of the Epon section. In three dimensions, therefore, the m e m b r a n e openings may be shaped like slits. The dashed arrow in Fig. 6 a points to a par t in the m e m b r a n e profile

which may be the edge of a hole, because this

spot in the m e m b r a n e is not entirely devoid of m e m b r a n e density. Third , i t is possible tha t the m e m b r a n e defects observed in this study are smaller than they really are; the glutaraldehyde may have caused a cer tain a m o u n t of shrinkage. Four th , the clusters of defects or holes would tend to go along wi th l ight microscope observations (11, 12) tha t some cells bulge in one region before hemolysing.

FIGURE 6 Showing abrupt discontinuities or holes in the membrane of the hemolysing erythrocyte when the cells are fixed by glutaraldehyde during the first minute of hypotonic hemolysis. The erythrocytes were added to hypotonie solution free of ferritin, and 12 see (Fig. 6 a and b) or 18 sec later (Fig. 6 c) glutaraldehyde was added. Tile cells were centrifuged, washed, and a solution of ferritin and colloidal gold was added. Membrane defects (arrows) were seen only during the first minute of hemolysis and only in those ghosts which contained ferritin (which appear as pale gray dots about 90 A in diameter) or gold (black dots). Ferritin-free ghosts (Figs. 6 b, c, and also 4 b) revealed intact membranes. The defects are between 200 and 500 A wide. The zone of membrane marked off by the bracket in Fig. 6 c probably does not contain a defect, the middle part of this piece of membrane having beer sectioned tangentially. These membrane defects occur in clusters, one cluster to a cell profile. The opening at the double-shafted arrow may be an edge of the "hole" or "pore" (see Discussion). Elmiskop I. Uranyl and lead citrate stains. Fig. 6 a, X 186,000; Fig. 6 b, X 88,000; Fig. 6 c, X 80,000.

66 THE JOURNAL OF CELL BIOLOGY • VOLUME 32, 1967

PHILIP SEEMAN Transient and Stable Holes in Erythrocyte Membrane 67

FIGURE 7 An example of bulging of tile cell membrane that occurs before the permeation of ferritin particles. These cells were fixed in glutaraldehyde 1~ sec after the onset of hypotonic hemolysis. Many cell profiles revealed a single bulge, as shown above; ferritin added to these fixed cells did not permeate them. These small bulges were never seen at later time points. RCA- EMU-3F. Uranyl and lead citrate stains. X ~4,000.

The Duration of Existence of the

Transient Holes

That most of the transient holes are open for approximately 10 to 15 sec is in reasonable agreement with the fading time of hemolysing erythrocytes (33). I t is possible that the 2-/~-wide streams of hemoglobin diffusing from hemolysing erythrocytes (12) may be a consequence of hemoglobin diffusing through the duster of tiny holes observed in the present study. By estimating the number and the size of the defects of the hemolysing cell, the diffusion coefficient of hemoglobin (34, 12) is calculated as 7.3 X l0 -7 cm2/sec (see Appendix). Polson (reference 37; see also references 38, 39) has found that the diffusion coefficient of hemoglobin is 6.9 X l0 -7 cm2/s ec. This agreement seems almost fortuitous, considering the uncertainty and range in the chosen values (see Appendix).

Are the Observed Defects the Same as Those Through Which the Hemoglobin has Been Released?

Considerable evidence has been presented in this paper that glutaraldehyde does not per se induce or create artifactual membrane defects. I t

has been explained, in connection with the hole- fixation experiments, that the membrane defects were observed only in cells which were fixed between approximately 10 and 20 sec after the onset of hemolysis and which were permeated by subsequently added ferritin. Membrane defects were never .found in a glutaraldehyde-fixed erythrocyte or ghost which was devoid offerritin. Cells which had been fixed by glutaraldehyde before the 10-see time point or after the 180-see time point were never permeable to added ferritin (Fig. 5), and the cell membranes never contained any defects. This indicates that glutaraldehyde does not of its own accord induce membrane defects.

The defects, therefore, must develop in association with the process of osmotic hemolysis. Sec- ondly, the holes are most frequently detected between 10 and 20 sec which is when the rate of hemoglobin release is highest. This fact further suggests that the transient holes are intimately associated with the hemolytic process. Third, the size and number of the transient holes are roughly adequate to account for the 10-see fading time in osmotic hemolysis (see reference 40), assuming the hemoglobin is released by diffusion.

For these three reasons it is likely that hemoglobin has been released through the observed defects which apparently admit the ferritin particles. I t is, of course, also possible that, while the "single- region openings" may have been the path of hemoglobin egress, there may be a second popula- tion of cells which rather develop "multiple-region openings" (with 150-A-wide holes) diffusely over the entire membrane (see also reference 41). Such cells could also be permeated by ferritin, but such small 150-A defects might not be detected in a 500 A-thick Epon section.

Membrane Pores Created by

Surface-Active Agents

The lysolecithin-treated cells have holes about 300-400 A wide in the plane of the thin section. The saponin ghosts do not show membrane discontinuities although ferritin (120 A in diameter), subsequently added, was admitted into the fixed ghosts; the membrane holes, therefore, must be smaller than the section thickness (500-600 A).

A P P E N D I X

The equation used for calculating the diffusion coefficient is (34, 12) D = ~ 7r r s In (Co/C)~ t 0.96 d N, where D is the diffusion coefficient of hemoglobin, % ~r r 3, the critical volume of the

68 THE JOURNAL OF CELL BIOLOGY " VOLUME 3~, 1967

FIGURE 8 Demonstrating the existence of membrane defects with prolonged patency after hemolysis induced by lysolecithin.

FIGVRE 8 a The cells were hemolysed in isotonic solution by lysolecithin, fixed in glutaraldehyde, washed, and then exposed to ferritin. The ferritin can be seen (as gray dots) on both sides of the cell membrane which appears scalloped. Elmiskop I. Uranyl and lead citrate stains. )< 64,000.

I~GURE 8 b High magnification of another cell from the experiment of Fig. 8 a. The membrane at the base of each indentation appears discontinuous. Although the indentations created by saponin are similar and although ferritin permeates the saponin ghost, no such membrane discontinuities were observed. Elmiskop I. Uranyl and lead citrate stains. )< 158,000.

cell jus t before the onset of hemolysis, Co, the cell hemoglobin concent ra t ion at the critical volume, C, the final concentra t ion in the superna tan t and the ghost (which are the same; reference 7), t, the fading time, d, the d iameter of an average hole (around 300 A), and N, the n u m b e r of holes per cell. The average length or d iameter of the zone or cluster of defects was 11,000 A; the n u m b e r of defects, occurr ing every 1800 A or so, within a circle of d iameter of 11,000 A is a round (11,000) 2 A2/(1,800) 2 A 2 or 37 holes. Since only one zone of holes was always seen per cell profile, it may be assumed tha t 37 (4 -50%) represents the holes per cell. The cri t ical volume is roughly 2 X 10 -1° cm 3 (35); this is 102% over normal (35)

R E F E R E N C E S

1. BARON, J. 1928. Arch Ges. Physiol. 220: 251. 2. ADAIR, G. S., J . BARCROFT, and A. V. BOCK. 1921.

J. Physiol., (London). 55: 332. 3. BAYLrSS, L. E. 1924. J. Physiol., (London). 59: 48. 4. TEORELL, T. 1952. or. Gen. Physiol. 35: 669. 5. HOFrMAN, J. F. 1962. J. Gen. Physiol. 45: 837. 6. GOURLEY, D. R. H. 1957. or. Appl. Physiol. 10:511. 7. HOFFMAN, J. F. 1958. J. Gen Physiol. 42: 9. 8. KLmANSKV, C., A. DE VRmS, and A. KAT-

CHALSKY. 1960. Pathol. Biol. 8: 2005. 9. TrmonoRn, J., and E. D. ROE:N. 1965. Clin. Res.

13 : 283. 10. I-[OFFMAN, J. F., J . I-hLLIER, I. J . WOLMAN, and

A. K. PARPART. 1956. J. Cellular Comp. Physiol. 47: 245.

11. DANON, D. 1961. J. Cellular Comp. Physiol. 57: 111. 12. KOCHEN, J. A. 1962. Am. J. Diseases Children. 104 :

537. 13. HAoois, G. H. 1965. J. Mol. Biol. 14: 598. 14. DOURMASHKIN, R. R., R. M. DOUOnERTY, and

R. J. C. HARRIS. 1962. Nature. 194: 1116. 15. BANGHAM, A. D., andR. W. HORNE. 1962. Nature.

196: 952. 16. GLAUERT, A. M., J. T. DINGLE, and J. A. LucY.

1962. Nature. 196: 953. 17. HussoN, F., and V. LUZZATL 1963. Nature. 197:

822. 18. PALADE, G. E. 1954, J. Exp. Med. 95" 285. 19. FARQUHAR, M. G., and G. E. PALADE. 1965.

J. Cell Biol. 26: 263. 20. KARNOVSKY, M . J . 1961. J. Biophys. Biochem. Cytol.

11: 729. 21. KATCHALSKY, A., O. KEDEM, C. KLIBANSKY, and

A. DE VRmS. 1960. in Flow Properties of Blood and Other Biological Systems, A. L. Copley

and will have a Co of abou t 17.5 g % (36). For the hole-development and the hole-fixation series of experiments the di lut ion of the packed erythrocytes was 1:50 in the hypotonic med ium; the final equi l ibr ium concentra t ion of hemoglobin, C, in bo th the ghosts and the superna tan t would be 35/50 g (see reference 36). For t, the fading time, we may choose a value of 10 see (see reference 40), which is roughly the same t ime tha t the t ransient holes remain open.

The author is grateful to Professor George E. Palade and Professor Walther Stoeckenius for their help and excellent suggestions.

Received for publication 5 July 1966.

and G. Stainsby, editors. Pergamon Press, N .Y. 155.

22. STOECr,~mUS, W. 1961. In Symposium of the Plasma Membrane. A. P. Fishman, editor. New York Heart Association, Inc., N. Y. 1066.

23. BAKER, R. F. 1964. J. Ultrastruct. Res. 11: 494. 24. BAYER, M. E. 1960. Arch. Ges. Physiol. 270: 323. 25. ROn~RTSON, J . D., 1959. Biochem. Soc. Syrup. (Cam-

bridge, Engl.). 16: 3. 26. TORMEY, J. McD. 1965. J. Cell Biol. 25: 1. 27. BRIOHTMAN, M. W. 1965. J. Cell Biol. 26: 99. 28. BRIGHTMAN, M. W. 1965. Am. J. Anat, 117: 193. 29. BONDAREF~, W. 1964. Exp. Neurol. 10: 377. 30. LEE, R. E., and J. D. FELDMAN. 1964. J. Cell

Biol. 23: 396. 31. ROSENBLUTH, J., and S. L. Wmsm. 1964. J. Cell

Biol. 23: 307. 32. H~Jxenv, H. E. 1964. Nature. 202: 1067. 33. PONDER, E. 1948. Hemolysis and Related Phe-

nomena, Grune & Stratton, N. Y. 34. FrucKn, It. 1934. J. Gen. Physiol. 18: 103. 35. RAND, R. P., and A. C. BURTON. 1963. J. Cellular

Comp. Physiol. 61: 245. 36. SAVITZ, D., V. W. SIDEL, and A. K, SOLOMON.

1964. 3". Gen. Physiol. 48: 79. 37. POLSON, O. 1939. KoUoid-Z. 87: 149. 38. ROUOHTON, F. J. w . 1959. Progr. Biophys. Biophys.

Chem. 9: 55. 39. KELLER, K. H., and S. K. FRIEDLANDER. 1966.

3". Gen. Physiol. 49: 681. 40. HEEDMAN, P. A. 1958. Exptl. Cell Res. 14: 9. 41. PARPART, A. K., and R. BALLENTINE. 1952. In

Modern Trends in Physiology and Biochem- istry. E. S. G. Barron, editor. Academic Press Inc., N. Y. 135.

70 T~E JOURNAL OF CELL BIOLOGY " VOLUME 8~, 1967

Documents

TRANSIENT HOLES IN THE ERYTHROCYTE MEMBRANE …